The appellant was convicted by a jury in the state district court of Barton County, Kansas, of burglary, larceny, forgery and two counts of uttering a forged instrument and sentenced to five concurrent thirty year terms under the state habitual criminal statute. On appeal, the Kansas Supreme Court sustained the conviction on two separate occasions, State v. Wood, 190 Kan. 778, 378 P.2d 536 (1963) and 197 Kan. 241, 416 P.2d 729 (Kan.1966). Appellant then sought a writ of habeas corpus in federal district court and was denied relief on the ground that he had not exhausted state remedies. After this court reversed and remanded for an evidentiary hearing,1 the district court determined that although the question was a close one, the search of the automobile appellant had been driving when arrested was not invalid, and even if it were, the admission of the evidence thereby obtained was harmless error.
The facts are not disputed and indicate that the Dr. Pepper Bottling Company in Great Bend, Kansas, was burglarized on December 29, 1961, and a typewriter, check protector and check book were stolen. The next evening, Sheriff Kline of Rice County, Kansas, was officially advised to watch for a certain described vehicle whose occupants had attempted to pass a check suspected of being one of those taken in the burglary. The sheriff proceeded west of Lyons, Kansas, where he observed the automobile appellant was driving, followed the vehicle into Lyons, and radioed his co-officer Samuels that he would stop the car at the edge of the city. The automobile was then stopped and the appellant was arrested as he emerged from the driver’s seat. Both the appellant and one Jesse Wyatt, a eodefendant in the state trial, were searched by officer Samuels who had joined in the arrest. The search uncovered a Dr. Pepper Bottling Company check in appel*396lant’s shirt pocket (later introduced in the trial as state’s exhibit 4). There was, however, no search of the automobile at that time.
Shortly after the appellant and Wyatt were taken to the county jail, Kansas Highway Patrol Trooper Murphy noticed the abandoned vehicle on the side of the highway. After communicating with the sheriff and obtaining the keys to the car, Murphy drove the vehicle to the jail where approximately twenty minutes after the appellant had been arrested, the vehicle was searched for the first time. It was then that Murphy discovered one signed check under the right front seat (state’s exhibit 3) and seventeen executed but unsigned checks together with seventy-five blank checks under the right front floor mat (state’s exhibits 1 and 2).
The State seeks to sustain Murphy’s search as being made incident to appellant’s arrest. Appellant counters by conceding that his arrest was valid, there being probable cause, but contending that the belated search of the vehicle was not incidental to the arrest, and absent a search warrant, violated the Fourth Amendment. He further asserts that the fruits of the unlawful search — some 93 checks — contributed to his conviction on each of the five counts, thus precluding an application of the harmless error rule.
Appellant relies primarily upon Preston v. United States, 376 U.S. 364, 84 S.Ct. 881, 11 L.Ed.2d 777 (1964), to sustain his contention that the search was not contemporaneous with his arrest. In Preston the search of the automobile of the defendants soon after it had been taken to a police garage and the defendants charged with vagrancy, was declared to be “too remote in time or place to have been made as incidental to the arrest * * 376 U.S. at 368, 84 S.Ct. at 884. . The Court observed that the justification for an incidental search — the need to protect the arresting officers, prevent escape, and preserve evidence — no longer applies after the vehicle and its occupants are safely committed to police control. “Once an accused is under arrest and in custody, then a search made at another place, without a warrant, is simply not incident to the arrest.” 376 U.S. at 367, 84 S.Ct. at 883.
We note, parenthetically, that both the search and the state criminal trial in the instant case occurred prior to the date of the Preston decision. There is, however, no need to be concerned with the propriety of a retroactive application of Preston inasmuch as the case merely reaffirmed principles previously announced by this court and others.2 “Even before * * * Preston * * * it was clear that a warrantless search of a car in police custody at a time after the occupants’ arrest and under circumstances where there is no danger of removal is illegal. * * * The Supreme Court’s holding to the same effect in Preston puts the matter to rest.” Smith v. United States, 118 U.S.App.D.C. 235, 335 F.2d 270, 273 (1964). Indeed, several courts, including this one,3 have applied Preston retroactively without discussing the problem, thereby giving at least implicit recognition to the fact that Preston did not represent a clear break with the past *397and therefore does not properly present a question of retroactivity.4
Clearly then, the holding in Preston is controlling unless it can be meaningfully distinguished from the case at bar. This the State attempts to do by relying on United States v. Dento, 382 F.2d 361 (3d Cir. 1967), where the search of the defendant’s vehicle after he had been arrested and the car driven to the police station, was upheld. The court distinguished Preston on two grounds: “First, unlike Preston the instant search was substantially contemporaneous with the arrest (the entire arrest and search procedure was completed within twenty minutes). Secondly, there was a reasonable nexus between the offense the defendant was arrested for and the search of his automobile, where as in Preston the search was entirely unrelated to the arrest on the charge of vagrancy.” 382 F.2d at 365-366. United States ex rel. Foose v. Rundle, 389 F.2d 54 (3d Cir. 1968), and Arwine v. Bannan, 346 F.2d 458 (6th Cir. 1965), also relied on by the State, purport to make the same distinctions.
This court cannot be persuaded by such artificial attempts at distinction without a difference. The Court in Preston in no way indicated that the time factor, i. e., the precise time period between arrest and search, was to be controlling. In fact, the Court did not indicate the amount of time involved except to note that the search was “soon after” the defendants had been booked at the station. Timing is significant only in the sense that the search is either at the time of the arrest or it is not; if it is not, it is irrelevant whether the elapsed time is twenty minutes or twenty hours. Rather, the crucial distinction, clearly indicated in Preston, is that if the search is not made at the time of the arrest, but is conducted at a later time when the vehicle is in custody and its occupants in jail, then the justification for an incidental search, e. g., the need to preserve evidence, protect the arresting officers, and prevent escape, no longer obtains and a warrant must be secured. Obviously, the absence or presence of a justification for the search will not depend solely upon whether the delay is brief or lengthy. Accordingly, it is the existence or nonexistence of a justification for a warrantless search — not the mere passage of time — that is determinative.
The second distinction attempted in Dentó, namely, analysis of the nature of the relationship between the offense prompting the arrest and the later search, can have significance in determining whether the search was reasonable only as it relates to probable cause. The court in Dentó must have intended to indicate that the subsequent search is valid when closely related to the reason for the arrest because when so related there is probable cause for the search. Cf., United States v. Butenko, 384 F.2d 554, 566 (3d Cir. 1967), vacated, 394 U.S. 165, 89 S.Ct. 961, 22 L.Ed.2d 176. The difficulty with that position is that it misconceives the function of probable cause in this context. Thus, the fact that officers have probable cause does not dispense with the requirement of a search warrant. Chapman v. United States, 365 U.S. 610, 81 S.Ct. 776, 5 L.Ed.2d 828 (1961). Consequently, it is not'the existence of probable cause, but the presence of factors justifying the failure to procure a warrant that is critical.5 *398When so viewed it becomes apparent that the relationship between the offense and the search is of no consequence since that relationship can have little bearing on the feasibility of first obtaining a search warrant.6
In brief, both distinctions propounded in Dentó overlook the fact that a search without a warrant is valid if incidental to an arrest only because the need to protect the officers, and other similar practical considerations, justify an exception to the general requirement that a warrant is a precondition to a valid search. When the justification is lacking, as in the case at bar where the vehicle was in police custody and its occupants in jail, then regardless of the time involved or the existence of probable cause for the search, a search without a warrant is simply not incident to the arrest.7 We must therefore conclude “that the search of the car without a warrant failed to meet the test of reasonableness under the Fourth Amendment, rendering the evidence obtained as a result of the search inadmissible.” Preston v. United States, 376 U.S. at 368, 84 S.Ct. at 884.
There remains for consideration the question of whether the admission of the unlawfully obtained evidence, state’s exhibits 1, 2 and 3, consisting of 93 checks in varying stages of completion, can be said to have constituted harmless error. To recapitulate, the appellant was charged and convicted on five felony counts: second degree burglary, grand larceny, forgery of state’s exhibit 4, and twice uttering a forged instrument (state’s exhibit 4). State’s exhibit 4, it will be recalled, was the check discovered in appellant’s shirt pocket, admittedly admissible and unchallenged by the appellant. This led the district court, adopting the reasoning in State v. Wood, 197 Kan. 241, 416 P.2d 729, 734 (1966), to determine that the forgery and two uttering counts were based exclusively on exhibit 4, with the inadmis*399sible evidence, exhibits 1, 2 and 3, being material only to the burglary and larceny counts. Reasoning that exhibit 4 was of equal materiality in obtaining the burglary and larceny convictions, the court concluded that the 93 checks were merely cumulative in nature, not the basis for appellant’s convictions, and not prejudicial to the substantial rights of the appellant. We must emphatically disagree.
Contrary to the approach taken by the district court, Fahy v. Connecticut, 375 U.S. 85, 86-87, 84 S.Ct. 229, 230, 11 L.Ed.2d 171 (1963), teaches that when considering a question of harmless error, “we are not concerned * * * with whether there was sufficient evidence on which the petitioner could have been convicted without the evidence complained of. The question is whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction.” Further elaboration in Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967), disclosed that “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.” The result is that before we find harmless error, we must conclude beyond a reasonable doubt that the 93 tainted checks did not contribute to appellant’s conviction.
As the case was submitted to the jury, exhibits 1, 2 and 3 were unrelated to the forgery and uttering counts since only exhibit 4 had been forged or uttered. The tainted evidence was supposed to be considered only in connection with the theft of the checks as charged in the burglary and larceny counts. As so considered, it is manifest that such evidence, comprising 93 of the 94 checks introduced, must have influenced the jury and contributed to the conviction.8 At the very least, it is clear that the State has not shown beyond a reasonable doubt that the inadmissible evidence was not a contributing factor.9 It follows that there was prejudicial error affecting the burglary and larceny convictions.10
Moreover, realism requires that relief be granted as to each of the three remaining counts. Admittedly, in the strict contemplation of law the invalid evidence should not have contributed to the forgery and uttering convictions. Yet in the minds of the jurors, appellant’s guilt or innocence of forgery and uttering would be affected by the initial conclusion on the burglary and larceny counts. The case must have appeared as one involving a theft, followed by an attempt to convert the stolen cheeks into cash; such was the clear implication urged by the State during the trial and cannot now be ignored. Consequently, the conclusion is inescapable that the tainted evidence might have contributed to the forgery and uttering convictions.
In sum, we conclude that the search of the automobile at the county jail after appellant had been incarcerated therein was in violation of the Fourth Amendment, and the admission into evidence of the fruits of that invalid search contaminated each of the five counts on which the appellant was convicted.
Reversed and remanded with directions to stay the issuance of the writ until the State has had a reasonable opportunity to retry the petitioner.